Mobile Tech & Gadgets

Huawei will manufacture its Kirin X90 for PC using the 7 Nm of SMIC

by

Huawei and Smic work side by side and intensity since they get it The US sanctions to China. Together they have managed to manufacture 7 Nm semiconductoras, for example, the soc Kirin 9000s of the Mate 60 Prousing an integrated circuit production technique known as Multiple patterning. This strategy in broad strokes consists in transferring the pattern to the wafer in several passes with the purpose of increasing the resolution of the lithographic process. His problem is that he usually has an upward impact on the cost of chips and the decline in production capacity, although it works. These two Chinese companies have been forced to use the Multiple patterning in your teams of deep ultraviolet lithography (UVP) because the prohibitions of the United States and the Netherlands prevent the Dutch company Asml from selling their machine their machines in Chinese extreme ultraviolet photolithography (UVE). The latter allow to produce more advanced semiconductors than UVP equipment. Huawei and SMIC seem blocked in the 7 nm lithography A little over a year ago, in February 2024, the newspaper Financial Times said having had access to the statements of two experts who defended that SMIC was finalizing the refinement of their semiconductor manufacturing processes in their UVP machines. Its purpose, according to this source, was to have the necessary technology to manufacture Integrated 5 nm circuits Massively before the year 2024. For SMIC and its clients, among which Huawei is, this would be a very important achievement. The triple folding of Huawei, one of its most ambitious proposals, reuses a 7 nm soc However, Huawei’s 5 NM socates have not yet appeared. The chip Kirin 9010 which has happened to the 9000S model is also being manufactured in the 7 NM SMIC node using the technique of Multiple patterning. We can find this soc, for example, in Huawei’s new triple foldingthe mate ultimate design, which We have tried Just two weeks ago at the Mobile World Congress in Barcelona. It is evident that the last smartphones of this company would have been wonderful to have a 5 -nm Soc. And to his GPU Asce for artificial intelligence (AI) too. But there is even more. According to Digitimes Asiathe CPU for PC Kirin X90 that will produce Huawei with the purpose of consolidating an alternative to Intel and AMD chips will also be manufactured in the 7 Nm node. In these circumstances it is reasonable to conclude that, for the moment, Huawei and SMIC are stuck in this integration technology. This is a big problem for China because it does not have the necessary technology to produce avant -garde semiconductors comparable to those who manufacture Intel, AMD or Qualcomm, among other western alignment companies. The great hope of the country led by Xi Jinping is having as soon as possible Your own UVE photolithography teamsand it seems that they are close. In fact, as we tell you last week, A filtration has revealed that Huawei is already testing a prototype of one of these machines. If this information is finally confirmed and China has its commercial UVE machines in 2026 will have taken a crucial step in its pulse with the US. Image | Hiilicon More information | Digitimes Asia In Xataka | TSMC acknowledges that it has been considered taking its factories out of Taiwan. It is impossible for a good reason

SoftBank will pay 6,500 million per ampere. A new war chapter for data centers has just been written

by

At the beginning of last February the interest of the Japanese Softbank investment group was made public in The acquisition of the chips designer American ampere computing LLC. This last company is specialized in the development of Processors for serverswhich already put at that time on the table SoftBank’s intention to expand your business in large data centers. It made sense in a context in which the rise of the artificial intelligence (AI) is promoting that these facilities multiply at full speed. Just a month and a half later that strategic movement has established itself. According to your own AmpereSoftBank has just closed your purchase. He will pay 6,500 million dollars for this company and will have one more letter in the presumably prosperous business of the data centers. This investment group is The Arm owner since 2016; At the beginning of last October Invested 500 million dollars in Openai; And, in addition, it is one of the companies that lead The Stargate program with which the US seeks to sustain its dominance of AI. The war for data centers for AI is already underway The large technology companies that are involved in the deployment of AI are facing multimillion -dollar investments to develop their data centers infrastructure. Microsoft has confirmed that 80,000 million dollars will be spent during the fiscal year of 2025. and Google, 75,000 million. On the other hand, the Stargate program budget that I have mentioned a few lines above rises to no less than 500,000 million dollars. This is the cake from which SoftBank is being able to seize. At least a good portion. Ampereone chips will reach 256 nuclei for 2025 and presumably have a very competitive energy efficiency However, to achieve it the companies that control, or control in the short term, such as ampere, whose processors are implemented on ARM technology, will have to Compete with Nvidia, Intel, AMD, Google or Amazon. In any case, the ampereone chips, which are being manufactured by TSMC in their 3 nm node, will reach 256 cores For 2025 and will presumably have a very competitive energy efficiency. These are his great buzas against the proposals of the competition. The current situation is triggering something that is worth not overlooking. As Dan O’Brien has observedthe president of the American Futurum consultant, technology companies are forging strategic alliances with the purpose of molding the industry to their measure and maximizing the economic performance of their investments. As we have seen, SoftBank is the owner of ARM. On the other hand, Oracle has a very significant participation in Ampere, which will soon belong to SoftBank, and which is ARM client. In addition, all the companies in which we have just repaired, SoftBank, ARM, Oracle and OpenAi, participate in the deployment of the infrastructure required by the Stargate program. In any case, This network of dependencies does not end here. As we have seen, SoftBank is an important OpenAI investor, and this last company is Oracle’s partner and is trying to develop His own chips for ia. Finally, to curl the curl further, rumors that defend that Oracle is interested in buying Tiktok, a company that belongs to the Chinese company bytedance persists. SoftBank also has an important participation in the latter. As O’Brien argues In his tweet “everything is connected (…) It is fascinating to see how the technology titans shape the industry.” Image | Ampere computing LLC More information | Ampere computing LLC In Xataka | The B300 GPU is the new Nvidia beast for Ia. And we already know what prepares for 2026 and 2027

The O1-PRO model of Chatgpt Pro is 140 times more expensive than the Chatgpt O3-mini. What is not clear is that it is much better

by

Paying $ 20 per month for accessing Chatgpt Plus did not seem too serious. The thing began to get interesting when Openai launched its chatgpt pro subscription at 200 dollars a month. But now there is an even more expensive version of these models, and this again shows that we are at the beginning of a worrying slope: that of AI prices. What happened. OpenAI He has just announced That O1-Pro, the most powerful version of its reasoning model, is now available through its API. Until now it was only possible to enjoy it in a limited way through the subscription to Chatgpt Pro, but now the developers can also create services with it. The striking is not really that the company offers this version, but the price at which it is available. 140 times more expensive than O3-mini. As we said, O1-Pro is only limited to Chatgpt Pro users. If someone wants to use it intensively through the API, prices They are really high: O3-mini O1 O1-PRO Price for 1m tokens input (dollars) 1.10 15 150 Price for 1m tokens output (dollars) 4.40 60 600 As can be seen, O1-Pro is 10 times more expensive than O1, and about 135 times more expensive than O3-mini, which is the version of the reasoning model that for example we can use (limitedly) in the free version of Chatgpt. Gold Price Tokens. That cost imposes an important barrier for those who want to use this AI model for projects in which we introduce a lot of text and want to obtain a lot of text. For example, if we want to analyze a lot of documents and We saturate The input context window (200,000 tokens) and the output (100,000 tokens), we will have a cost of $ 90 … for a single question and an answer (an API request). But is it really so good? Preliminary O1-Pro impressions in Chatgpt Pro They were not especially striking according to whom They have tried italso They have protested for one falling quality. According to Openai O1-PRO, it offers “responses that are better consistently”, and highlight that it is an O1 version that uses more calculation power to “think more and offer even better responses to the most complex problems.” For asking not to be. IA companies are losing a money training their models, but they hope to recover it with a simple technique: increasingly faces. Data leaked in The Information also pointed out how Openai were already raising how an AI agent could replace a “human doctorate” to investigate, the cost would be $ 20,000 per month. They seem faces, but maybe they don’t. Of course, the price of these subscriptions and these accesses to the API may seem expensive, but they can also be a true bargain. What matters here is the profitability that we can get out of these models of AI: if they help us generate many more income, we will end up having “robotic employees” who work 24 hours a day, seven days a week and also do so with a productivity and quality of theoretically extraordinary work. But of course, we will first have to use them like this … and ensure that these results are effectively valid, something that for now does not seem so clear. In Xataka | AI agents are promising. But as in Tesla’s FSD, you better not take your hands from the steering wheel

streaming has generated a Milmillionaire bleeding in Apple accounts

by

Since, a few months ago, Apple started taking Expenditure containment measures related to its audiovisual productions it was learned that the company not only had to abandon its project to become a Major that competed with Disney or Warner in its same terms. In addition, the expense generated by Apple TV+ was excessive. But today we have known that it goes far beyond what the most pessimistic calculations ventured: it loses one billion dollars a year. It is not another Netflix. Today, In an interview for Variety commemorating his 25 years at the head of Netflix, Ted Sarandos responded thus when asked about his impressions about Apple’s role in the streaming: “I don’t understand it beyond a marketing play, but they are very intelligent people. They may see something we don’t.” Saraonds’s words can be an order or not, but it is clear who carries the lead: in July, Bloomberg spoke Of the immediate cuts that Apple TV+ was going to start because it had less vision in a month than Netflix in one day. The abyss between both competitors is indisputable. One billion a month. The Information Publish a report that analyzes the financial statement of the platform after five years. It states that Apple TV+ loses more than one billion dollars a year. The specialized medium in Apple 9to5Mac Comment on the articlestating that although it had always been said that the platform was not profitable, it had never had such concrete and forceful figures associated with its losses. A lot of investment, few income. The explanation of these spectacular figures is at the high cost of Apple films in recent years. The very expensive ‘Napoleon‘by Ridley Scott and’The Moon Killer‘Martin Scorsese were the first, but the authentic financial bomb was’ argylle’, which raised around 35 million dollars And it cost 200. After that, to a commercial candy like ‘Wolfs‘, starring Brad Pitt and George Clooney, predicted a storm, and was released directly on the platform, with Apple assuming losses that perhaps they would have been greater carrying it to cinemas. The thrust of ‘separation’. The funny thing is that, at the image level, Apple TV+ goes through an excellent time. ‘Separation‘It has passed, with its long -awaited second season, of being a product of cult to reveal itself as one of the most unanimously acclaimed series of the moment. It is said that in the last month he could have reported two million users to the platform, which would be added to the 45 that The Information affirmed that he had at the end of 2024. And on the horizon, the return of ‘Ted Lasso‘. Apple is going well. However, these certainly disastrous numbers for the streaming They do not imply that the company does wrong. The last quarter, Apple closed the year with 124.3 billion dollars of incomeof which 36,300 could be considered benefits. Despite some obstacles, such as the fall in device sales in China, the business works perfectly for Apple; Even subscriptions, section where Apple TV+is, but also Apple Music, App Store, ICloud or Apple Care, paid great, growing 14% compared to the previous year. That is, bad news for Apple TV+, but they are certainly far from making the colossus staggered. Header | Apple In Xataka | Prime video is becoming an “aggregator” of other platforms. And the arrival of Apple TV+ is the last example

The Google Pixel 9 has dropped (again) price after the announcement of Pixel 9a. Thus one of the best 2024 mobiles stay

by

Yesterday the Google Pixel 9abecause of the PCComponentes, it has decided to increase the discount that the Google Pixel 9. Until a few hours ago, we could find it for 699 euros instead of the official price of 899 euros, but now we have even cheaper: the store has it right now by 653 euros. * Some price may have changed from the last review A mobile that still has many years of Android updates He Google Pixel 9 It was (and it still is) One of the best mobiles that were launched in 2024. At the level of specifications we are faced with the most complete mobile in its official price range, even more taking into account that it is on offer. Mount a very good screen Super acts OLED of 6.3 inches which offers a 120 Hz soda rate and that is compatible with HDR10+. Its processor is the Google Tensioner G4is accompanied by 12 GB of RAM and 128 GB of internal storage and has a good assortment of artificial intelligence functions through Gemini. Regarding its photographic section, its main camera (from the rear cameras module) of 50 MP (82º, F/1.68, OIS) stands out mainly. It also has a good 4,700 mAh battery that supports 45W fast charging, 18W wireless load and reverse wireless load and Your operating system will receive updates up to six years. You may also interest these accessories for the Google Pixel 9 Jetech Magnetic case for Google Pixel 9/9 Pro 2024, compatible with MAGSAFE, Translucent Antigolpe Housing Matte Backup (Black) * Some price may have changed from the last review Google Pixel Buds Pro 2 – Wireless button headphones with active noise cancellation – Bluetooth headphones – Obsidian * Some price may have changed from the last review Some of the links of this article are affiliated and can report a benefit to Xataka. In case of non -availability, offers may vary. Images | Pepu ricca In Xataka Android, Google In Xataka | The best mobiles, we have tried them and here are their analysis In Xataka | The best price quality price. Your analysis and videos are here

That solar energy stops growing in Spain

by

This month will be remembered Like the rainiest, having water reserves in 65.8%, According to the miteco. However, it is not good news for solar energy that has fallen for the first time in three years. The data. According to the reports of the Spanish Electricity (REE), in the month of March, the Solar Generation He has reached 11% total, which represents a decrease with respect to the year 2024 (13.9%) and The year 2023 (12.8%). Heavy rains. We have had a couple of weeks that it is difficult to see the sun and it seems that You can extend even more. For this reason, one of the most harmed clean sources has been photovoltaic. However, other renewables have come out more than beneficiaries of this rain and wind, we are talking about wind and hydroelectric. Favored systems. On the one hand, wind energy It does not have the same capacity that photovoltaic, but when it comes to adverse weather conditions who eats the ground to solar panels are wind turbines. In fact, Ree’s real time data shows that wind is generating 46% wind energy compared to 12% of solar (These data may vary, since they are being taken when writing this article). On the other hand, as we mentioned above the total water reserve, this situation has allowed hydroelectric energy to have reached in what we have been 17.5%. However, it has not yet exceeded last year that achieved a level 21.5% of electric production, more than four higher percentage points. So how will my light bill be? Although solar energy always It had been key to the energy mixthis time the contribution of both wind and hydroelectric can be A small relief for consumers. However, like has denounced the OCU There are other factors that will make the light bill, such as The price of gas and electric tolls. Image | Flickr Xataka | Solar panels compared to 100,000 olive trees: in Jaén the neighbors are also protesting against renewable projects

Like ‘the Sims’, ‘Inzoi’ simulates your real life in a video game. The difference is that it takes it very seriously

by

Within just over a week, the life simulator ‘Inzoi‘It will reach computers around the world. The first impressions about the same They are landing And all speak of the impressive graphic effort that involves their recreation of the everyday, as well as the differences and similarities with the other milestone of the genre, ‘THE SIMS‘. However, its characteristics mainly derived from its origin, South Korea, raise many other doubts about their acceptance beyond the Asian market. 20 years without life simulators. ‘The Sims’ is one of the most famous video games of all time, among other things because it dazzled an important public sector that did not usually play, long before the term casual It became fashionable with the advent of mobile devices. However, the last main installment of the game, ‘The Sims 4’, was launched in 2014. With a Fifth Discarded delivery and the transformation of the franchise into a free-to-playits original model is exhausted. Does it make sense that there is a game that intends to revalidate the throne of life simulators? The big difference. To say that ‘The Sims’ is a game that aims to replicate everyday life is the same as saying that’The Simpsons‘It is a reliable portrait of an American family. Or, for not getting out of video games, which ‘Simcity‘He is a realistic urban manager. Soft or more strident, everything is portraits softened by satire, because in Western society we are unable to look at ourselves without the filter of cynicism and self -parody (or, in the case of ‘Simcity’, of bleeding criticism). That is the big difference with ‘Inzoi’, which takes very seriously its replica of real life. Such is life. In fact, that is the main criticism that is falling to ‘Inzoi’: “Not having understood“That ‘The Sims’ that supposedly imitates, between the scandalous realism of Unreal Engine and the possibility of climbing our own gestures, physical and movements. Or that complex system of reactions that makes our relationship evolve with the rest of the characters as we behave with them, in a maze of social ties that impacts the city’s own growth and the map. An amusingly ambitious concept that distances itself from ‘The Sims’, which more than a life simulator was a mischievous omnipotence: put your SIM in the pool and remove the stairs, to see what happens. Pure Korea. Possibly, the greatest “problem” to which Inzoi will face when penetrating in non -Asian countries, once the widespread glare passes before their graphics, it is his spirit so typical of Asian simulators and role -playing games: the need to take it seriously. Of course, in ‘Inzoi’ we can do the goat as in a ‘GTA’, but a karma system will protect the digital universe from excessive riot, which is what the European or American player tends: to force the limits of the games and simulation. There is nothing more to see a game of ‘GTA online’ to understand to what extent is the simulation of very different shapes between opposite points of the globe. Our simulator is social networks. Don’t we like simulation outside of China or Korea? No: rather We replace it with social networksthat cover our aspirational and voyeurist needs. We do not build complex and digital replicas of ourselves in video games, because that is our Twitter, Bluesky and Instagram profiles, where we perform a life that we want to pretend or that we long to carry, often raising it in a harmless and playful way. The greatest obstacle before him ‘Inzoi’ is that, without knowing it, we have already created the perfect simulation. And without the need for latest generation computers. In Xataka | ‘Inzoi’ wanted to be the new ‘The Sims’, but the crisis that social simulators live has also ended their expectations

Social Security has published the work of work casualties in 2024 and we have bad news: we have broken a record

by

Spain is living A worrying phenomenon: Labor casualties have reached a new historical maximum in 2024, marking a trend that marks its fourth year up and does not seem to be. Demographic aging, saturation of health or improving employment figures are the main suspects. Record in work casualties due to illness. The increase in Low for temporary disability It is not only a perception of entrepreneurs, but a reality backed by Official security data Social of 2024 that has just been published. According to The published by The countrySocial Security records reveal that the average prevalence of temporary disabilities for common contingencies (those not directly related to the work) reached its highest historical percentage: 53.3 per 1,000 employees in 2024. That of autonomous professionals is 39.8 per 1,000 employees. A high cost for the economy. According to the report ‘Labor absenteeism for common contingencies’ Prepared by the employer Asedas and the association of mutual accidents at work (AMAT), in Spain in 2024 8,716,663 casualties for diseases and non -work accidents occurred, which meant a cost of more than 28,987 million euros in social security benefits of social security and direct cost for companies. This expense has increased by 99% in the last six years. The average time that a worker remains medical leave due to temporary disability is 39.2 days by professional contingencies (those directly related to work activity), while for self -employed professionals it increases to 106.5 days. The aging of the workforce. One of the main reasons behind this increase in medical casualties due to common contingencies is the general aging of the population and, with it, a greater tendency to get sick. According to data from the Employers of Mutual Amat, more than half of the work casualties occur in the 31 to 51 -year section. The INE data They point out that more than 35% of workers in Spain are 50 years or older and ‘Study on the aging of the templates in the General State Administration and future projection at 2030’ It points to that in 2023, the peak of the demographic pyramid will reach the strip between 50 and 60 years. According to data of The independentif we compare the current 50 and 60 -year -old strip with that of 2002, it has increased by 169%, while the segment of between 16 and 24 years has fallen 36%, and the one aged 25 to 34 has done 20%. The collapse of public health. The worsening of public health also directly affects the resolution of the casualties due to temporary disability, since, in many occasions, these casualties are lengthened more than necessary by waiting lists in diagnostic tests and surgical interventions. The official data of the Ministry of Health of December 2023 (the last available) indicate that Spain registered a record of 849,535 people in surgical waiting lists. This figure represents almost twice as many people in this comparative situation With the data of 20 years ago. The average waiting time for surgery (in 2023) was 128 days, which practically double the 77 days of 2003. Mutuals can help. The increase in labor casualties not only affects workers, but also a significant blow to the Companies productivity. Unions, employers and social security They have reached agreements so that mutuals have a greater prominence with the aim of decongesting the waiting lists of public health. This public-private collaboration frame is intensified mainly in casualties due to temporary disability with traumatic pathologies, in which mutuals achieve expedite tests and treatments necessary reducing the medical leave time and saving several billions of euros to companies and social security. In Xataka | The last idea of ​​the government in labor matters: a “flexible” decrease that allows at the same time to work In Xataka | Cádiz’s ghost official: six years without going to work and charging his payroll. They caught him when they were going to give him a prize Image | Social Security, Unspash (Owen Beard)

snatch the AI ​​market to Nvidia

by

Nvidia leads the GPU market to artificial intelligence (AI) with a shocking forcefulness, especially if we stick to training the great language models. According to the consultant Jon Pedie Research The company led by Jensen Huang controlled 90% of the market During the third quarter of 2024, and presumably their numbers will not have changed much since then. AMD reached a 10%quota during that period, so the presence of Intel in this sector is anecdotal. The sanctions that it has deployed The US government During the last two and a half years they prevent Nvidia from selling its GPU for the most advanced, such as chips H100, A100, H800 or A800. However, the GPU that at the time the US Department of Commerce does allow you to sell its Chinese clients is the H20 cut chip. And he is having a lot of success. In fact, its sales in the country led by Xi Jinping grow 50% quarter to quarter despite the delicate current situation. AMD wants a cake portion in China The Chinese government wants its technology companies to stop buying integrated circuits to NVIDIA. The step he took at the beginning of last October was merely informative, but, in any case, It was expressly addressed to companies that acquire GPU for AI with the purpose of recommending that they buy chips produced in China, and not the solutions of the Jensen Huang company. The current situation does not seem ideal for another American company to try to grow in the Chinese market Nvidia is losing competitiveness in China despite the good sales of her H20 chip. It is inevitable. And this GPU on paper is much less capable than The most sophisticated chips for the that Jensen Huang’s firm is currently selling. At the moment the Ministry of Industry and Information Technology of China has not officialized anything, but the administration led by Xi Jinping has restricted since last August the purchase of NVIDIA H20 GPUs by Chinese companies. On the other hand, Huawei has its own GPU for ia, The chips ascend tofor more than five years. During this period of time it has been refining them and increasing its capacities with the purpose of matching or even overcoming the performance of the A100 and H100 chips of Nvidia. This situation does not seem ideal for another American company to try to grow in the Chinese market, but this panorama does not seem to have discouraged Lisa her, the general director of AMD. And this executive is now in China. In fact, according to SCMP He has begun his visit by highlighting the role that his company is playing in the development of AI models of the country’s main actors, among which are Deepseek and Alibaba. Smooth his has remarked that THE INSTICT GPU of AMD are compatible with the models of AI of these last two companies in a clear attempt to Defend your position in the Chinese market. Besides, has pointed out which has observed a perceptible increase in the performance of Depseek models thanks to AMD hardware optimization. This statement is quite surprising if we take into account that Depseek uses Nvidia chips in training and huawei in inference. Whatever it will be interesting to verify how AMD hardware evolves for the Chinese market during the next months. Image | AMD More information | SCMP In Xataka | The Nvidia pulse and US administration becomes more virulent. The B20 GPUs for danger

bet on the Open Source

by

China with the Open Source models It is a zasca throughout the mouth For the unbelievations. It didn’t seem logical that the Asian giant Take it like this With the development of artificial intelligence systems open and available for all audiencesbut the truth is that in recent months the country has given a unique lesson the world. He has good reasons for this. Above all, taking into account that its great rival, the USA, has adopted a totally opposite position. Except in outstanding exceptions as a goal with flame “That the Chinese models probably inspired,” the large technology companies and startups that try to win the AI ​​race do so with a closed and owner approach. Ours is ours: You can use it, but we are not going to tell you (much) how they work, they seem to tell us. But in recent weeks China has not stopped demonstrating that another way of doing things is possible. And do them well. Alibaba, Baidu or Tencent have been flooding the market for some time with the Open Source models, Something simply unusual in a country known for its secrecy and its obsession with control. Until the appearance of Deepseek Everything seemed to follow the expected script in the war between the US and China. The first did not stop imposing vetoes and zancadillas on the Chinese technology industry. Nothing to export advanced chips of AI or to suggest the secrets of our technology. China was expected to come with secret, closed and owner developments, but the Deepseek moment marked a turning point. Or two. The first, demonstrating that China was already at a very high level in AI. The second, by betting on an Open Source approach and invited the world to try Deepseek freely and learn from it. Since then they have not stopped appearing from China that they presume to have greater performance than Deepseek R1 and that of course make it clear that in China there is an entire internal war to lead this segment. Qwen 2.5 Max (Alibaba), Doubou 1.5 Pro (Bytedance), Ernie 4.5 (Baidu), Hunyuan Turbo s (Tencent) or Kimi K1.5 (MoNshot AI) are some recent examples of that frantic movement, but again the really surprising thing is the open approach of many of these models. And as they point out In Financial Times“business and geopolitics, Generosity rarely lacks strategy“That China is opening its AI models has its reason for being, and that is precisely to attract talent to strengthen its models and thus be able to win that career. In China they are managing the US sanctions decentralizing their development and managing to capture the interest of developers around the world to study these models and improve them. In fact, that overall opening of Chinese models suddenly makes vetoes to use Nvidia advanced chips are no longer so effective. It is true that in China they will not be able to use them, but Western countries with access to these chips can train and refine models based on Chinese Open Source models, which also serves as a way of avoiding sanctions. It is a “no matter that you do not let us train our models with the best Nvidia chips: others will do it outside China.” The strategy seems to be working. The interest and popularity of Chinese models is growing, but it is also their quality and benefits is already at the level of the best closed openai, Anthropic or Google models. And if there are open models with the same performance as chatgpt or Claude or Gemini, but they are source and free, cheaper or cheaper, Why use payment alternatives? In many cases the answer will be overwhelming: not at all. Of course, the Chinese bet has its risks. Foreign companies may end up taking advantage of Chinese advances to launch their own models that are superior or that fit other markets. Perplexity, for example, has created R1 1776an Open Source version of Deepseek R1 that annuls censorship and biases imposed by some Chinese models – they don’t answer you if you ask about the events of Tiananmen’s Square. It is something that the Beijing government probably did not like it, but it was inevitable. And despite this, China is achieving what it pursued. And the solution was not to close the world more, but, curiously, open to the Open Source philosophy. Who would have said it. In Xataka | Deepseek has become the Fashion company of the AI. Unlike the rest, he refuses to accept external investment

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.